When Does Balanced Matter?

[March Audio]

And I have thought it is a scientific forum.

Well my professional background of 30 years is instrumentation and measurement systems. I think the results of that test are on WBF somewhere.

 

[pirad]

Well my professional background of 30 years is instrumentation and measurement systems. I think the results of that test are on WBF somewhere.

Please describe the "speaker cable effect" you measured , explain methodology and results and provide corroboration in literature and/or peer conducted experiments.

 

[SIY]

Phono balanced is an excellent way to go- with my MC setup, I can actually touch a cartridge pin with my finger and not get hum.

The SE vs balanced noise question is somewhat complicated because there's several ways to achieve balanced, and the answer is different. For example, in my MC preamp ("His Master's Noise"), the balanced input is a transformer with a high input CMR, which steps up the signal by 20dB to the input of a single-ended amplification stage. Hence, there's essentially no noise penalty for balanced (the transformer's Johnson noise is negligible compared to the input tube). In my MM stage ("The Equal Opportunity"), the input is active differential, so there's a 3dB noise penalty. But (and there's always a but), that 3dB isn't reflected in the actual system performance because the input stage noise is well below the Johnson noise of the cartridge+load, so ends up being completely, errrrr, lost in the noise.

 

[RayDunzl]

I really like your avatar.

It catches my eye every time I see it, and makes me laugh.

 

[March Audio]

Please describe the "speaker cable effect" you measured , explain methodology and results and provide corroboration in literature and/or peer conducted experiments.

No. I dont care if you dont believe me. I am not going to waste my time on this because you are miffed that I contradicted your statements. Suffice to say that the frequency response alters dependant upon cable properties. Well no shit Sherlock Im not sure why you would think that is a contentious result. Basic AC theory explains it.

 

[watchnerd]

Well to my mind yes, certaonly in the sense that like a microphone, it could take advantage of a differential amp input and the benefits of its common mode rejection.

Being from an instrumentation background, I am somewhat bemused why you wouldn't want to use a differential phono amp.

The major impediment is lack of phono stages with balanced inputs.

In my case, the phono stage on my Devialet Expert is an ADC, and supposedly has differential wiring, but the sockets are RCA instead of XLR.

 

[March Audio]

The major impediment is lack of phono stages with balanced inputs.

In my case, the phono stage on my Devialet Expert is an ADC, and supposedly has differential wiring, but the sockets are RCA instead of XLR.

Yes I agree that is the obvious place to use a diff input. You will still get some CMR with that arrangement but its not ideal by any means.

 

[SIY]

One of the advantages of doing your own designs and builds is that you don't have to worry about mixing and matching random commercial equipment, you can do an optimized system design. My next step, after I get this fershlugginer SP10 Mk2 turntable working again, is to build an instrumentation amp directly into the turntable plinth using some very cool AD8229 chips I got from Scott Wurcer (late of Analog Devices).

 

[SIY]

I really like your avatar.

It catches my eye every time I see it, and makes me laugh.

View attachment 12431

It very much was my childhood.

 

[pirad]

No. I dont care if you dont believe me. I am not going to waste my time on this because you are miffed that I contradicted your statements. Suffice to say that the frequency response alters dependant upon cable properties. Well no shit Sherlock Im not sure why you would think that is a contentious result. Basic AC theory explains it.

Fair dinkum. Said like a true scientist and a gentleman.

 

[Kal Rubinson]

One of the advantages of doing your own designs and builds is that you don't have to worry about mixing and matching random commercial equipment, you can do an optimized system design. My next step, after I get this fershlugginer SP10 Mk2 turntable working again, is to build an instrumentation amp directly into the turntable plinth using some very cool AD8229 chips I got from Scott Wurcer (late of Analog Devices).

The last phonostage I built was based on an INA103 (http://www.ti.com/lit/ds/symlink/ina103.pdf) for balanced input. A delight.

 

[SIY]

Yeah, that's a really excellent choice for a chip to use there.

 

[DonH56]

Seems like more (in)fighting than anything else around here lately...

Random thoughts:

I advocate short speaker wires and long interconnects but rarely worry about it. My speakers specify <0.07 ohm source cable resistance for the crossover to work properly. You need 12-AWG wires for runs of about 20' or so. Other speakers have lower impedance dips so arguably need shorter runs. And of course an amplifier with higher output impedance will change what cables you need (though tube amps are so high it probably doesn't really matter).

I have run into amps that have issues with stability and/or RFI when using long speaker cables (usually RF noise coupled on the lines that gets back to the input through the feedback path where it can be rectified and amplified back out to the speakers). I have run RCA connections to about 50' (probably longer, don't really remember) without significant signal loss but ground loops and noise can be a killer. IME the biggest advantage of balanced connections for consumers is to obviate the ground loops, followed by noise rejection since some live near big noise sources (outside the house, like radio/TV stations or industrial motors and such, or inside, like dimmers and motors in the house).

There are AES papers talking about speaker runs and RFI, and balanced vs. single-ended runs, but I am no longer a member so can't cite sources. Which makes this all hearsay, of course. Whatever.

The INA103 and many similar instrumentation amps are great with low source impedances and loads like MC and strain-gauge cartridges, but can be noisy with MM cartridges because the current input noise is fairly high (leading to higher noise for e.g. a 47 k-ohm MM load). That is a typical trade for these guys; the best low-noise design must take source impedance into account. The INA103 also has fairly low gain (20 dB by default IIRC) which is probably OK as a pre-preamp but may be a little low as a do-it-all phono preamp.

There have been some balanced phono stages through the years though I can't remember which ones off-hand. There was another thread here or on AVS and a few were named. Not all phono cartridges are completely balanced though most are "close" if not there; the way the mechanical assembly is performed also plays into it IIRC. Some TT's used multi-pin DIN connectors to provide balanced outputs; I don't recall seeing a TT with XLRs but there probably were some.

The cost of balanced is relatively high when talking large numbers of units. The circuits cost more, the connectors cost more and take up a lot more panel and board space, etc. I'd expect cost pressure and general lack of demand and/or need to relegate balanced circuits to high-end products.

FWIWFM - Don

 

[SIY]

There's a few cartridges that have a grounding strap (some of the Shures). Cut that and you're balanced, at least up to high enough frequencies that the few pF of strays are negligible.

Fortunately, cartridges with high source impedance tend to have high output, so there's no significant penalty to buffering with a low noise jFET if needed.

edit: Regarding connectors, the micro XLR from REAN take up no more space than an RCA and are similarly inexpensive.

 

[Fitzcaraldo215]

Seems like more (in)fighting than anything else around here lately...

Random thoughts:

I advocate short speaker wires and long interconnects but rarely worry about it. My speakers specify <0.07 ohm source cable resistance for the crossover to work properly. You need 12-AWG wires for runs of about 20' or so. Other speakers have lower impedance dips so arguably need shorter runs. And of course an amplifier with higher output impedance will change what cables you need (though tube amps are so high it probably doesn't really matter).

I have run into amps that have issues with stability and/or RFI when using long speaker cables (usually RF noise coupled on the lines that gets back to the input through the feedback path where it can be rectified and amplified back out to the speakers). I have run RCA connections to about 50' (probably longer, don't really remember) without significant signal loss but ground loops and noise can be a killer. IME the biggest advantage of balanced connections for consumers is to obviate the ground loops, followed by noise rejection since some live near big noise sources (outside the house, like radio/TV stations or industrial motors and such, or inside, like dimmers and motors in the house).

There are AES papers talking about speaker runs and RFI, and balanced vs. single-ended runs, but I am no longer a member so can't cite sources. Which makes this all hearsay, of course. Whatever.

The INA103 and many similar instrumentation amps are great with low source impedances and loads like MC and strain-gauge cartridges, but can be noisy with MM cartridges because the current input noise is fairly high (leading to higher noise for e.g. a 47 k-ohm MM load). That is a typical trade for these guys; the best low-noise design must take source impedance into account. The INA103 also has fairly low gain (20 dB by default IIRC) which is probably OK as a pre-preamp but may be a little low as a do-it-all phono preamp.

There have been some balanced phono stages through the years though I can't remember which ones off-hand. There was another thread here or on AVS and a few were named. Not all phono cartridges are completely balanced though most are "close" if not there; the way the mechanical assembly is performed also plays into it IIRC. Some TT's used multi-pin DIN connectors to provide balanced outputs; I don't recall seeing a TT with XLRs but there probably were some.

The cost of balanced is relatively high when talking large numbers of units. The circuits cost more, the connectors cost more and take up a lot more panel and board space, etc. I'd expect cost pressure and general lack of demand and/or need to relegate balanced circuits to high-end products.

FWIWFM - Don

Excellent advice, as always, Don.

The issue maybe gets a little complicated by the distinction between true differential balanced electronics and the fake balanced that appears in most consumer audio, like HT systems, except for the costly upper tiers.

Honestly, I have no idea what the status is in pro audio electronics. Yes, the cables are usually balanced as are the mics, but are the electronics always true differential balanced? How much does it really matter?

Ok, confession. All my stuff is fake balanced as far as I know. Fortunately, that results in substantially lower cost than differential balanced, but it also still provides the same freedom from EMI/RFI external noise and ground loops as does fake balanced. The "penalty" is running the single-ended signal through the extra internal inverting circuitry, these days probably primarily an op amp, AFAIK. I do not know if the old theoretical advantage of differential balanced still holds, based apparently on times past when transformers were predominant in that task.

I cannot say that I have meticulously done DBTs on this. I just do it. It sounds fine to me, and I just love working with the XLR connectors. So, no sleep lost.

 

[DonH56]

I am not sure what you mean by "fake balanced". There are several differential and quasi-differential topologies that I have seen widely used though have not done any sort of exhaustive survey recently. Here are a few off-the-cuff:

1. Fully differential active circuits, e.g. instrumentation amplifiers or differential pairs of devices;
2. Transformer-coupled differential circuits, which may tie the "other" side to ground to perform differential-to-single-ended conversion (either way);
3. Quasi-differential outputs using an inverter on one side, e.g. two op-amps with one configured as non-inverting and the other inverting to derive a differential input;
4. Quasi-differential inputs using a single op-amp (not an instrumentation amplifier) to perform differential-to-single-ended conversion (usually has poorer common-mode rejection ratio) -- could also do this with two op-amps like the output circuit above but summing the op-amp outputs to pass on to single-ended circuitry within the component;
5. Impedance-balanced quasi-differential by using a resistor to ground to "float" the (-) side so looking into the input or output the DC resistance (and AC impedance to a point) is the "same".

I am sure I am forgetting a few but those are the main ones I recall off-hand. I have seen all of these in pro and consumer components. I do think transformer circuits are declining due to cost, size, and potential for hum pickup but am not up on the latest high-end sound boards.

HTH - Don

 

[Kal Rubinson]

The INA103 and many similar instrumentation amps are great with low source impedances and loads like MC and strain-gauge cartridges, but can be noisy with MM cartridges because the current input noise is fairly high (leading to higher noise for e.g. a 47 k-ohm MM load). That is a typical trade for these guys; the best low-noise design must take source impedance into account. The INA103 also has fairly low gain (20 dB by default IIRC) which is probably OK as a pre-preamp but may be a little low as a do-it-all phono preamp.

Yup. My application was with a Koetsu Black Gold-Line (now you know how far back this goes) and the INA103 pre-pre fed a 47K-input RIAA/gain stage based on a Burwen audio opamp module. The old days.

 

[Fitzcaraldo215]

I am not sure what you mean by "fake balanced". There are several differential and quasi-differential topologies that I have seen widely used though have not done any sort of exhaustive survey recently. Here are a few off-the-cuff:

1. Fully differential active circuits, e.g. instrumentation amplifiers or differential pairs of devices;
2. Transformer-coupled differential circuits, which may tie the "other" side to ground to perform differential-to-single-ended conversion (either way);
3. Quasi-differential outputs using an inverter on one side, e.g. two op-amps with one configured as non-inverting and the other inverting to derive a differential input;
4. Quasi-differential inputs using a single op-amp (not an instrumentation amplifier) to perform differential-to-single-ended conversion (usually has poorer common-mode rejection ratio) -- could also do this with two op-amps like the output circuit above but summing the op-amp outputs to pass on to single-ended circuitry within the component;
5. Impedance-balanced quasi-differential by using a resistor to ground to "float" the (-) side so looking into the input or output the DC resistance (and AC impedance to a point) is the "same".

I am sure I am forgetting a few but those are the main ones I recall off-hand. I have seen all of these in pro and consumer components. I do think transformer circuits are declining due to cost, size, and potential for hum pickup but am not up on the latest high-end sound boards.

HTH - Don

By "fake" balanced I mean, of course, everything except your item 1., all of which require added circuitry in the signal path, as is common. Added circuitry in the signal path has been the theoretical "no-no" of critics of balanced in analog circuits, unless done by fully active (duplicated) circuits.

 

[pirad]

What Floyd Toole is to sound reproduction, Bill Whitlock is to balanced audio. See the definitive material on balanced from Whitlock on the exquisite Rod Elliott site.
http://sound.whsites.net/articles/balanced-2.htm
For those with little time here's the skinny: pay attention to CMRR. If it's not given, chances are the common mode rejection ratio is substandard or even detrimental to the system's overall sound quality. Best CMRRs are offered by transformers. A good Jensen is 120dB or more. The next in line is Whitlocks inGenius chip licensed to That Corp., CMRR 90dB . Then....various solutions with figures of 70,60 and even less.
Rod Elliott's foreward and conclusion are also very informative. If you are to read only one chapter from Whitlock's paper look at the "Design Checklist". It is really not trivial to design a good balanced system.

 

[DonH56]

By "fake" balanced I mean, of course, everything except your item 1., all of which require added circuitry in the signal path, as is common. Added circuitry in the signal path has been the theoretical "no-no" of critics of balanced in analog circuits, unless done by fully active (duplicated) circuits.

Hmmm... I'd disagree but not worth debating. Transformers vary but good ones offer really good common-mode noise rejection and are certainly "balanced". They can also provide complete galvanic isolation, something most active circuits do not, so I would not rule them out. In fact most of those schemes do a decent job, though the simplest resistor-balancing scheme (#5 in my list) generally offers pretty poor rejection, partly because the resistors are not usually tightly-matched, and partly because the resistor won't balance the op-amp's rising output impedance with frequency. But, even those (used by inexpensive pro mixers as well as some consumer gear) can provide some isolation from ground (shield) noise.

"Added circuitry" in audio circuits has always been the bane of some audiophiles. Heaven help them if they ever see the actual recording chain.